A JAX Research Toolkit for Visualizing, Manipulating, and Understanding Gemma Models with Multi-modal Support based on Penzai.
-
Updated
Jan 13, 2026 - Jupyter Notebook
#
gemma-scope
Here are 4 public repositories matching this topic...
A mechanistic interpretability workstation that maps neuron activations and polysemantic features across LLaMA and GPT-2 weights using Sparse Autoencoders (SAEs) and TransformerLens. Patch-isolates the Indirect Object Identification circuit to verify linear Taylor approximations through an interactive dashboard. (326 characters)
postgresql telemetry python3 pytorch circuit-analysis sparse-autoencoders ai-safety sae asyncpg linear-probing ai-alignment fastapi mechanistic-interpretability representation-engineering activation-steering gemma-scope transformer-lens
-
Updated
Aug 7, 2026 - JavaScript
A research tool for studying how deception emerges in multi-agent LLM systems and detecting it through activation analysis.
alignment gemma sparse-autoencoders multi-agent-systems ai-safety emergent-behavior interpretability deception-detection activation-analysis mechanistic-interpretability llm-agents gemma-2b gemma-scope transformer-lens linear-probes
-
Updated
Jan 11, 2026 - Python
Chemistry interpretability pilots on Gemma-2-2B via Neuronpedia — SAE feature discovery, causal steering, and circuit tracing. Preliminary evidence for a mechanistic-interpretability predoctoral proposal (no local GPU).
chemistry gemma sparse-autoencoders interpretability llm mechanistic-interpretability activation-steering gemma-scope circuit-tracing neuronpedia
-
Updated
Jul 6, 2026 - Python
Add this topic to your repo
To associate your repository with the gemma-scope topic, visit your repo's landing page and select "manage topics."